do keep in mind that it is somewhere between difficult and impossible for one person to both do modifications to sound gear and choose which result is better where subtle differences are involved.
I certainly think there are challenges even for an experienced builder in doing modifications. I think much of this is simply a matter of personal taste and subjectivity. I believe "better" is usually subjective far more then not.
I think amp building is a blend of science/math and art (broadly defined). I think if an experienced and somewhat knowledgeable builder does one modification at a time and listens to it over some time and A/B's the mod to the original (especially at different volume and tone settings and environments), that even subtle modifications are reasonable efforts and can yield positive desired results. I think it's possible to have a reasonable measure of
predictability around using a specific modfication.
If someone implements a modification and posts a schematic of that. Then others implement the same modification and then report similar results (or post soundclips indicating similar results), I think that adds a measure of "validity" to the modification (validity meaning it will yield similar results). Part of "emperical based research" is that someone else duplicates the findings of the original research and gains similar results.
I would argue that some entire amps are a collection of subtle modifications that went brought together as a whole are no longer subtle. Howard Dumble perhaps took a Fender amp and started modifying it. At one point, his schematic of this modified amp is discovered and others reproduce it and get similar results. That doesn't mean everyone will think a Dumble amp is "better", but it can indicate that using those modifications on a Fender amp may have a reasonable measure of predictability.
[On the D'Mars ODS schematic I posted earlier in this thread, I am willing to venture that perhaps one out of 4 or 5 modifications I left in & the rest of them I removed which finally yielded the posted schematic. The original idea was to borrow aspects of Marshall and Dumble topology and try to blend them together in a cathode biased amp where I could switch from 6V6 to 5881's. And I wasn't aware of anything quite like that out there, so I built an idea and began a lengthy process of modifying it one mod at a time. Now others have built the amp and gotten similar results per sound clips that are posted. On the Hoffman forum, I've seen others duplicate the So-Low Watt, HoSo56, Tweed Overdrive Special and Tweed BluezMeister & get similar results per report and/or soundclips. All of those were ideas first built in one form and then a length process of modifying the original idea took place]
Does attempting modifications have significant challenges? Yes. Can they achieve similar desired results with predictability when implemented within a reasonably similar situation? Yes. Does it make the amp "better"? The answer to that is very subjective.
I believe even in the field of applied science, there is a significant "human factor" involved.
I say all of the above respectfully and hope it is received that way. With respect, 10thtx
that's $220.56 at todays exchange rate 
